A non-trivial number of adolescents and young adults suffer from an eating disorder, many of which lead to serious health problems. Therefore, creating a validated classification scheme is of major importance, as it is needed to help guide treatment and prevention programs. Among adolescent and young adult women in the United States and the United Kingdom, approximately 0.1-1% have anorexia nervosa and 1-3% have bulimia nervosa. Although the Diagnostic and Statistical Manual for Mental Disorders-IV (DSM-IV) criteria are widely used to classify individuals, the eating disorder diagnoses were not empirically defined and the majority of eating disordered individuals do not meet the criteria for anorexia nervosa or bulimia nervosa. Moreover, the diagnostic criteria may not capture all early onset cases and it is unclear whether they are appropriate to use with males. We propose to collect additional data on eating disorders and treatment from two ongoing cohort studies, the Growing Up Today Study (GUTS) in the United States and the Avon Longitudinal Study of Parents and Children (ALSPAC) in the United Kingdom to empirically create an eating disorder classification scheme. We will use latent class analysis and compare the resulting classification to the existing DSM-IV classification, as well as another classification proposed from latent class analyses in two cross-sectional studies. We will use data from the 16,882 participants in GUTS who have been followed every 12-24 months since 1996 to empirically derive an eating disorder classification scheme using latent class analysis. We will determine the best groupings of eating disorder symptoms, as well as the best cut-offs (such as engaging in binge eating > once per week vs. > two times per week). We will use comorbidity and outcome (weight gain and weight status, comorbid depression, anxiety, substance and alcohol use, and persistence or worsening of symptoms) during adolescence and young adulthood as the validators. By collecting information on treatment history we can assess whether these patterns and associations vary by gender, age, or treatment history in the GUTS cohort. We will use the same eating disorder information collected at ages 14, 16, and 18 years in ALSPAC to test whether the observed results in GUTS can be replicated and are therefore generalizable. We propose to use the two largest prospective studies with frequent assessments of disordered eating during adolescence and young adulthood to empirically derive a classification scheme and compare it to the system that is already frequently used (DSM-IV), but is known to not work well for many individuals, as well as an alternative classification system that has been proposed. The size of our sample and the repeat assessments allows us not only to empirically determine the best groupings of symptoms, but also allows us to propose valid frequency cutoffs for binge eating and purging.